A PSO-optimized novel PID neural network model for temperature control of jacketed CSTR: design, simulation, and a comparative study.

This paper proposes a particle swarm optimization (PSO) tuned novel proportional integral derivative (PID) like neural network (PSO-PID-NN), to control the temperature of a nonlinear jacketed continuous stirred tank reactor (CSTR). The nonlinear continuous stirred tank reactor (CSTR) plant is one of the most popular reactors in the chemical industry. The proposed structure is elegant in design, having only three neurons in the hidden layer and a single output neuron. The three weights in the neural network's output layer represent the PID controller's proportional, integral, and derivative gains. The suggested approach uses the PSO method to optimize the output layer weights, which corresponds to the PID gains. Mean square error is used as an objective function to optimize the weights. The performance of the proposed PSO-PID-NN controller is tested by comparing the time domain specifications of the output response, against the conventional Zeigler Nichols tuned PID controller and the back propagation-based NN-PID controller (BP-NN-PID). The overshoot in the proposed controller is 23.13%, while it is 26.33% in BP-NN-PID, and 44.13% in Zeigler Nichols tuned PID controller. In addition, the rise time is 0.1283 s, while it is 0.2727 s in the BP-NN-PID controller and 0.2813 s in Zeigler Nichols tuned PID controller. The proposed controller is also tested for disturbance rejection, it was found to be more efficient in rejecting disturbance signals as compared to BP-NN-PID and ZN-tuned PID controllers. [ABSTRACT FROM AUTHOR]